Jingbin Wang

Ph.D Candidate, Computer Science Dept, Boston University

Biography

I am a research assistant in Image&Video Computing Group, and my academic advisor is Margrit Betke. I defended my Ph.D thesis on Jan 23rd, 2007. My thesis title is Object Segmentation Using Shape Constraints. Since Feb. 2007, I have worked for Google, Inc. at Mountain View, CA.

M.S in Computer Science& Engineering, Zhejiang University, Hangzhou, China,2001

B.S in Computer Science&Engineering, Zhejiang University, Hangzhou, China,1998

Research

My current research focuses on object segmentation and recognition in computer vision. Please refer to my resume for further information.

Shape-based Object Segmentation and Recognition:  
Incorporating Shape Prior into Discrete Optimization Methods, 2005--present.
Detecting Object Classes with Variable Shape Structure, 2005--present.
MosaicShape: Stochastic Region Grouping with Shape Prior, 2004--2005.
Shape-based Curve Growing with Adaptive Regularization, 2003--2004

Medical Image Analysis:  
Medical Image Analysis Interface, 2001--2004
Lung Contour Segmentation / Surface Registration, 2001--2002.
Pulmonary Fissure Segmentation and Lung Lobe Visualization, 2003--2004.

My former research mainly focused on computer graphics and solid modeling. Please refer to my old resume for knowing more about my past.

Selected Publications

Journal Papers:

[J1] Detecting Objects of Variable Shape Structure with Hidden State Shape Models,  Jingbin Wang, Vassilis Athitsos, Stan Sclaroff, Margrit Betke, IEEE Trans. on Pattern Analysis and Machine Intelligence, 2007. To Appear.

[J2] Pulmonary Fissure Segmentation on CT,  Jingbin Wang, Margrit Betke, Jane P. Ko, Medical Image Analysis, (10) 530-547, 2006. pdf

[J3] Segmentation of Nodules on Chest Computed Tomography for Growth Assessment,  William Mullally, Margrit Betke, Jingbin Wang, Jane P. Ko, Medical Physics, 31(4) 839-848, April 2004.

[J4] Embedding Visual Cognition in 3D Reconstruction from Multi-view Engineering Drawings,  Weidong Geng, Jingbin Wang, Yiyang Zhang, Yunhe Pan, Computer-Aided Design, (34) 321-336, Elsevier Science, 2002.  pdf

Conference Papers & Technical Reports:

[C1] Object Detection at the Optimal Scale with Hidden State Shape Models, Jingbin Wang, Vassilis Athitsos, Stan Sclaroff, Margrit Betke, BU-CS Technical Report BUCS-TR-2006-026, Computer Science Department, Boston University, 2006. NEW A robust hand tracker is implemented based on this paper. Here are some demos: demo1, demo2, demo3

[C1] Detecting Instances of Shape Classes that Exhibit Variable Structure, Vassilis Athitsos, Jingbin Wang, Stan Sclaroff, Margrit Betke, 9th European Conference on Computer Vision (ECCV), Graz, Austria, 2006. (paper and poster)

[C2] MosaicShape: Stochastic Region Grouping with Shape Prior, Jingbin Wang, Erdan Gu, Margrit Betke, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), San Diego, U.S.A, 2005. (pdf, talk slides)

[C3] Generating Sequence of Eye Fixations Using Decision-theoretic Attention Model, Erdan Gu, Jingbin Wang, Norm Badler, 3rd International Workshop on Attention and Performance in Computational Vision (WAPCV), in conjuction with CVPR, San Diego, U.S.A, 2005.

[C4] Shape-based curve growing model and adaptive regularization for pulmonary fissure segmentation in CT, Jingbin Wang, Margrit Betke, Jane P. Ko, 7th International conference on medical image computing and computer assisted intervention (MICCAI), pp 541-548, Saint-Malo, France, 2004. (pdf)

[C5] An intergrated chest CT image analysis system-"BU-MIA", Margrit Betke, Jingbin Wang, Jane P. Ko, Technical Report 2004-001, Computer Science Department, Boston University, 2004.

Other Publications

Programming

I am an experienced coder but maybe not a good programmer indeed. The follows are some collections among those I have developed in the past years.

This integrated program implemented two of my recent works[C1, C2], in which we applied the model-based approaches for object recognition purpose.

This integrated system was developed for our project of medical image analysis. Its main functionalities include pulmonary structure detection, recognition and registration. The details of this system was described in our work[J2,C4,C5].

This program was developed during my summer internship at GE Global Research Center in 2005. It aims for segmenting vascular trees from lung parenchyma on chest CT.

This cookie program was implemented based on a published SIGGRAPH work, which performs the colorization of grayscale images.

This cookie program was implemented based on another published SIGGRAPH work. Click here for more detailed description and results.

This was one of the oldest programs I had written before. It performs 3D solid reconstruction based on the multiple 2D projection engineering draws. Please refer one of my former work[J4] for more details.

This cookie program was initially implemented for CS585 as a programming project assignment. It applies the active contour technique to localize the object boundary.

Favorites